Research On Design Of An Underactuated Hand And Grasp Planning

The robotic hand is of vital importance in the interaction with its surroundings.As for radiation source disposal robots working in the unstructured environment,the key factor to complete a task is to grasp and handle the radioactive source successfully.With a number of driving units,the full-drive robotic hands are difficult to be anti-radiative.By comparison,the underactuated hands cannot grasp objects accurately because of their poor flexibility of fingers.In an effort to solve the above contradiction,the thesis designs a special mechanism to make the hand adapt to the shape of an object and improve the finger flexibility.The optimization,structural design and other related researches are then carried out.In this theses,the robotic hand grasping model and the object model are established,and the structural parameters which need to be optimized are presented.Stability index,position sensitivity index,transmission ratio index and task isotropic index are introduced to assess the grasping quality.Then,a non-dimensional comprehensive performance index which integrates all above independent indices is proposed to evaluate the grasping quality.Structural parameters are optimized using MATLAB robot toolbox and the optimal algorithm for the robotic hand,which provides a theoretical foundation for the hand structural design.Based on above optimization results and design requirements,this thesis presents the structure design of the robotic hand.The functional requirements of the hand are analyzed,and the basic configuration is determined.Then the functional blocks,such as the finger module and the motor driving module,are designed on the basis of the modular design idea.Additionally,in order to compensate for lacking flexibility,a novel memory alloy-based joint locking mechanism is present ed to improve the hand grasping ability.Finally,key components are selected by proper calculation.According to the structural characteristics of the hand and the requirements of grasping tasks,six grasp planning methods for conventional objects are developed.In order to expend the range of application,a fingertip rotating pinch planning method is developed for a small size target which cannot be successfully picked up by conventional ones.The driving circuit is designed based on the Elmo driving module.On the basis of the Beck Hoff controller and Ether CAT bus communication,the experimental platform for the underactuated robotic hand is built up.The output force of each knuckle and the maximum grasping force are calibrated by a series of experiments with the hand mounted on the UR manipulator.Then the effectiveness of the proposed joint locking mechanism and the grasp planning methods designed for different grasp targets are verified.Through grasping the objects of various sizes,shapes and masses,the grasping flexibility and the shape adaptability of robotic hand are then demonstrated.